Crosswind Sensitivity of Passenger Cars and Influence of Chassis
1. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
Crosswind Sensitivity of Passenger Cars and the Influence of Chassis
Ram Bansal1, R. B. Sharma 2
1,
Research Scholar, Depart ment of Automobile Engineering,RJIT BSF ACEDEM Y Tekanpur
2,
Hod of Mechanical Engineering department, RJIT BSF A CEDEM Y Tekanpur
Abstract:
Results of vehicle crosswind research involving both full-scale driver-vehicle tests and associated analyses
are presented. The paper focuses on experimental crosswind testing of several different vehicle configurations and a
group of seven drivers. Atest procedure, which utilized wind-generating fans arranged in alternating directions to
provide a crosswind "gauntlet", is introduced and described. Driver preferences for certain basic chassis and
aerodynamic properties aredemonstrated and linked to elementary system responses measured during the crosswind
gauntlet tests. Based on these experimental findings and confirming analytical results, a two-stage vehicle design
process is then recommended for pred icting and analysing the crosswind sensitivity of a particular vehicle or new
design.
Keywords: (A) vehicle dynamics considerations (e.g., weight distribution, tire, and suspension characteristics), (B)
vehicle aerodynamic properties, (C) steering system characteristics (most notably steering system comp liance, friction
and torque assists), and (D) driver closed-loop steering behaviour.
1. Introduction
This paper is based on recent findings fro m a vehicle aerodynamics research project [1]sponsored by the
Chrysler Motors Corporation at the University of Michigan Transportation Research Institute. The general thrust of
that research was directed at the crosswind sensitivity of passenger cars, and specifically, the influence and interaction
of chassis characteristics. The key elements considered in that study are outlined in Figure 1 and included: (A) vehicle
dynamics considerations (e.g., weight distribution, tire, and suspension characteristics), (B) vehicle aerodynamic
properties, (C) steering system characteristics (most notably steering system comp liance, frict ion and torque assists),
and (D) driver closed-loop steering behaviour and preferences obtained from experimental crosswind tests. This paper
focuses on the experimental crosswind testing conducted during the project using several different vehicle
configurations and a group of seven drivers. It reports on driver preferences for certain basic chassis and aerodynamic
properties and demonstrates a linkage to elementary system responses measured during those crosswind tests.
The paper begins with an examination of previous research findings related tocrosswind sensitivity of
passenger cars. A computer model, developed under this research program, is then briefly described. The use of that
model at different stages of the research helped to identify and probe certain vehicle-related mechanisms identified as
possibly significant contributors to the issue of crosswind sensitivity. As will be seen, the ability of the model to
predict basic dynamic behaviour patterns, observed in experimental measure ments of driver-vehicle systems operating
in crosswinds, is an important factor for recommending it as a tool within the vehicle design process. The basis of the
conclusions of this paper, however, rest upon experimental measurements and evaluations of a group of seven test
drivers operating seven distinctly different vehicle configurations during nearly identical crosswind conditions. The
crosswind tests were conducted using a set of eight fans (arranged in an alternating direction over a length of several
hundred feet) to approximate a rando m-like crosswind "gauntlet" driven repeatedly by each driver.
Fig. 1. Key Elements in Vehicle Crosswind Stability.
||Issn 2250-3005(online)|| ||January || 2013 Page 95
2. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
2. Previous Research
The trends during the last decade or so to lighter, more fuel efficient cars in response to changing energy
policies, co mb ined with more recent trends toward higher performance passenger cars, have led to increased interest in
aerodynamic styling as a means for providing low drag configurations and for mitigating any high-speed crosswind
sensitivities. In many cases, attempts at streamlin ing passenger cars for minimizing drag have led to unwanted
increases in crosswind sensitivity. As noted in such comprehensive texts such as Hucho [2]and Scibor-Rylski [3],this
tradeoff was observed as early as 1933 by Kamm [4]out of which arose the well-known truncated rear-end design
("Kamm-back") wh ich helped to offset much of the crosswind susceptibility introduced from streamlin ing. More
recent observations, such as Kohri and Kataoka [5]or Hucho [2],have contributed to improved understandings on the
subtle influences relating to A-and C-p illar styling designs and their importance in affecting crosswind sensitivity of
passenger cars. Studies such as Noguchi [6]have also noted the importanceof certain suspension properties (e.g., ro ll
steer and lateral force co mpliances) as elements not to be discounted when considering a v ehicle's crosswind
sensitivity. Other studies relating to crosswind sensitivity of passenger cars have been compiled in such documents as
Kobayashi and Kitoh [7],which was primarily concerned with literature related to the crosswind sensitivity of light
weight cars.
Wind Measurement Technol ogy
In addition to the studies cited above, a number of technological advancements in on-board wind
measurement capabilit ies have also occurred in the last two decades which have helped to promote further
understanding of the nature of crosswinds acting on vehicles in open areas, as well as in the vicinity of fixed roadway
objects and other moving vehicles. Smith in 1972 [8] described a MIRA wind measurement device (utilizing
anemo meters) and its use in measuring a number of d ifferent crosswind profiles. More recently, Tran [9] presented a
pressure transducer system for measuring the wind forces and moments acting on a vehicle by recording the pressures
at approximately 10 points along the circumference of a vehicle and then combining this informat ion with a uniform
flow model. A wind transducer developed by the Chrysler Corporation aerodynamics department utilizing a strain-
gauged sphere in combination with an inertial mass compensating accelero meter, is des cribed by Pointer in [10]. This
latter device was also used during the crosswind testing described subsequently in this paper.
3. A Driver-Vehicle Crosswind Model
A computer model for predicting the interaction between vehicle and driver during crosswind conditions is
introduced and outlined briefly in this section. Its use as an advanced tool that can be used in place of, o r in
conjunction with, certain types of dynamic random crosswind test procedures is subsequently being recommended as
part of a total crosswind sensitivity design process. The model was developed during this research and was used to
help identify and separate different mechanisms of the driver-vehicle system contributing significantly to the
crosswind sensitivity of the system. Figure 2shows a block diagram outlining the principal co mponents of the
crosswind driver/vehicle model. The technical details of the computer model can be found in Sayers et a1 [11].The
four primary co mponents of the model are described briefly in the fo llowing.
Vehicle Model
The vehicle model is characterized by five degrees of freedo m for the sprung mass, constant forward speed,
and massless suspension/wheel assemblies. Tire and suspension compliances are also included. Tire lateral fo rce is
treated as largely linear except for cornering stiffness dependency on vertical load. The basic dynamics of the vehicle
are very similar to that developed by Segel [12].High speed test data, collected during the course of the aerodynamic
crosswind stability research program at UMTRI [1],were used to validate the baseline model behaviour through direct
comparisons with model predictions. Test track handling measurements and aerodynamic wind tunnel measurements
of a baseline passenger car, in several different aerodynamic configurat ions, were conducted at nominal speeds of 50
and 100 mph to validate the model. As part of the model validation process, a stable platform and a variety of
transducers were used to measure all body motions. Steering wheel displacement and torque measu rements, front
wheel rotations, and other steering system functions (power boost pressures, pitman arm motion, and tie rod forces)
were also recorded and utilized in the model validation.
||Issn 2250-3005(online)|| ||January || 2013 Page 96
3. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
H front wheel angle contribution fro m steering co mmand and co mpliance
C front wheel angle contribution fro m suspension kinematics
f w total front wheel steer angle
Xf non-aerodynamic vehicle response
Fig. 2. Chrysler/UMTRI Crosswind Vehicle Model.
4. Experimental Crosswind Measurements
The results of experimental crosswind tests of driver-vehicle systems conducted during the research program
are presented in this section. All full-scale testing was performed on the vehicle dynamics facility at the Chrysler
Proving Grounds in Chelsea, Michigan. The vehicle crosswind tes ting results were obtained with the use of eight U.S.
Govern ment-owned wind generating fans described in detail by Klein and Jex in [13].
The use of fan-generated crosswinds in this research program was based upon several needs. One need was to
obtain crosswinds of sufficient magnitude that the test drivers would be given a definite subjective impression of the
different vehicle configurations being driven, and, that would also produce a system response that could be readily
measured by on-board instrumentation. Test speeds of 90-100 mph were selected so that the aerodynamic inputs
provided by sizeable crosswinds (25 mph or so) would still permit a linear-regime characterizat ion of the vehicle
aerodynamics. A second need was to have available a generally repeatable set of crosswind conditions so that different
drivers could be exposed to more or less the same crosswind experiences at widely vary ing times during the test
program. And finally, there was the practical need to be able to schedule test drivers on a regular basis and be
guaranteed that sufficient crosswind test conditions would be available.
Against this back-drop of perceived needs was the clear observation, wellreported in the literature, that fan -
generated crosswinds were generally insufficient for obtaining reliable subjective evaluations from test drivers. Since
nearly all of the previous uses of fans for subjective evaluations were for short duration drive-by scenarios, in which
drivers attempted to regulate the lateral path in response to a short-term pulse of crosswind, a different approach was
considered for this research program. It was decided that the traditional, closely-grouped fan arrangement, which
provides a short-duration pulse of crosswind, would be used only to evaluate the passive (non-driver) vehicle response.
Anew arrangement of the fans, in the form of a crosswind "gauntlet" course, would be used, ins tead, to expose the test
drivers to the fan-generated crosswinds over a longer period of time for co llecting their subjective evaluations.
Test Procedures
Two basic test maneuverers were used to evaluate the response of the various driver-vehicle configurations.
The first maneuverer was a fixed steering wheel drive-by of a constant pulse of fan-generated crosswind. It was used
to primarily characterize and verify differences in the passive crosswind behaviour of the different vehicle
configurations. The second maneuverer, serving to evaluate the active (driver and vehicle) crosswind system
behaviour, employed the same fans spread out over a longer distance but arranged in alternating directions. Active
steering for path regulation by each driver was required during this latter test. All tests were conducted for vehicle
speeds of 90-100 mph. Further descriptions of these two test procedures follow.
*Crosswind Pulse. This maneuverer was conducted with the eight fan units grouped togetherand facing perpendicular
to the test track as seen in Figure 3. Fixed steering wheel drive-by tests were then performed for each vehicle
configuration to evaluate their respective passive (no regulatory driver steering) wind behaviour. Each test v ehicle was
driven in a straight line at 100 mph fro m the amb ient environ ment past the fans whereupon it encountered an
approximately constant 25 mph crosswind for a period of nearly 0.7 seconds. A pulse -like vehicle response due to
entering and then exiting the crosswind stream was recorded by on-board instrumentation. The resulting peak yaw rate
and lateral accelerat ion responses observed for each vehicle configurat ion were then used to confirm that significant
and distinctly measurable differences in the vehicle aerodynamic andchassis configurations were
Vehicle trajectory
||Issn 2250-3005(online)|| ||January || 2013 Page 97
4. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
Fig. 3. Crosswind Pulse, Fixed Steering Wheel Test.
present in the passive crosswind behaviour of eachvehicle. (The short duration of the crosswind pulse at these speeds
did not permitthe vehicle to fu lly establish itself in a steady-state turning condition.)
*Crosswind "Gauntlet" Maneuvre. For th is set of tests, the eight fan units were located, in an alternating manner,
along opposite sides of a straight-line testcourse (single lane width) and distributed over a longitudinal travel distance
of approximately 350 feet. See Figure 4. Th is arrangement presented the driver-vehicle system with a series of
fluctuating pulses of crosswind fro m one side and then the other in a repeating sequence. The spacing between the fans
was approximately 52 feet. W ind output from each fan was set at an approximate level o f 25 mph and the test course
was driven at speeds of 90 to 100 mph. An inner lane width of approximately 8 to 9 feet was defined by a series of
traffic cones along the centre of the course in order to require each driver to regulate the vehicle path (without undue
demand) within those bounds during traversal of the crosswind course. All of the subjective evaluations collected
during the test program were obtained fro m this crosswind test procedure.
Impressions of several drivers who drove past the fans in both arrangements(Crosswind Pulse versus the
Crosswind Gauntlet) noted significant differences. The closely grouped pulse arrangement had a very s mall effect on a
driver's subjective and objective response as the cross wind pulse was encountered. The primary inputs to the drivers
were reported to be sound (fro m wind noise) and a mild change in direction as the fans were pass ed. Drivers also
commented that the experience was too brief. In contrast, the driving experience through the crosswind gauntlet
generally made a much'stronger impression on the test drivers. This was most likely due to the longer length of time of
crosswind exposure provided by the gauntlet course and significantly increased driver -vehicle system responses during
this test. For the same levels of fan wind output, the gauntlet course produced noticeably amp lified system responses
compared to those obtained from simple drive-bys of the closely-spaced pulse arrangement. The alternating pulses and
their input frequency during the gauntlet test produced a more resonating dynamic response that further contributed to
subjective driver impressions. The on-board measurements, as well as simple v isual observation of the different
vehicle configurations travers ing the gauntlet, confirmed the amplifying qualit ies of the gauntlet test procedure. As an
example, in a few of the worst-case vehicle configurations, peak lateral accelerations greater than 0.6g were recorded
at the driver head position and more than 0.45g's on the stabilized platform. Levels approximately half these would
have been recorded in the crosswind pulse test for the same fan output. Drivers Seven different drivers were ut ilized in
the test program to provide subjective evaluations of each of the vehicle configurations during the crosswind
gauntlettests. Objective measurements of the total driver-vehicle system responses werealso collected for four of these
test drivers to obtain representative system responses during the gauntlet tests. All of the drivers were males with
backgrounds as engineers or technicians ranging in age from 25 to 55. All drivers, but one, were associated to varying
degree with the crosswind research program. (Other drivers also participated intermittently during the testing but were
not included in these results because of not having driven each of the different vehicle configurations, or, because their
chosen speeds fell significantly outside the nominal 90-100 mph range.)
||Issn 2250-3005(online)|| ||January || 2013 Page 98
5. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
Fig. 4. Crosswind Gauntlet Test Course Used in Study
5. Conclusions
The results of the full-scale driver-vehicle crosswind tests presented above, in combination with static turning
analyses of vehicles in constant crosswinds and more comp lete dynamic crosswind simulat ions, suggest the following
conclusions:
A vehicle's static turning response due to a constant crosswind input and fixed steering wheel angle is a useful, first -
stage predictor of driver's likely subjective evaluation of a vehicle's crosswind sensitivity.
That same static turning measure will also frequently predict a vehicle's likely ranking of RMS responses obtained
during dynamic crosswind maneuvers.
A more reliable and accurate method for predicting subjective evaluations ofvehicle crosswind sensitivity is with
RMS yaw rate values obtained from fullscale testing or comp rehensive dynamic simulat ion of driver-vehicle
systems during dynamic, random-like or natural crosswind conditions.
Increased roll motion due to decreased suspension roll stiffness was associated with lower d river subjective
evaluations of vehicle crosswind sensitivity.
At vehicle speeds of 90-100 mph, variat ions in fore-aft weight distributionplayed as important a role as comparable
variations in aerodynamic canter-of-pressurelocation in influencing both subjective and objective evaluations
ofvehicle crosswind sensitivity.
A two-stage vehicle design process is recommended for analysing the crosswind sensitivity of a potential vehicle or
new design: (1) A preliminaryscreening of candidate vehicle designs for crosswind sensitivity, based uponthe
simp lified statics analysis of Equations (3) or (4),should first be conductedto screen out ineligib le candidate designs
having unsuitable vehicle properties(e.g., relative locations of mass canter, neutral steer point, and aerodynamicCP
that promote high passive crosswind sensitivity). (2) conduct a more in-depthand comprehensive analysis of the
"final round" candidate designs usingdynamic analysis (such as the crosswind model described above). The
dynamicmodel should employ random-like, natural crosswind inputs to examine likelydriver-vehicle responses to
systematic variat ions of vehicle chassis properties(suspension, steering system, and weight distributions particularly)
and differentaerodynamic designs. RMS values of system responses (e.g., yaw rate) can beused to evaluate the
influence of alternate designs.
||Issn 2250-3005(online)|| ||January || 2013 Page 99
6. I nternational Journal Of Computational Engineering Research (ijceronline.com) Vol. 3 Issue. 1
The dynamic crosswind model developed under this work can be used to further explain and analyse the
crosswind sensitivity of driver-vehicle systems in a dynamic context, part icularly for driving scenarios involving
active driver steering control during representative random crosswind conditions. Its use as a tool to
systematically examine the influences of vehicle sub-components on crosswind stability is especially useful.
Further man-machine research into likely driver p references regarding body roll motion and driver-cantered
motion experiences deriv ing fro m aerodynamic crosswind forces and mo ments is recommended.
Re-examination by other parties of the crosswind gauntlet test procedure (orsimilar procedures) utilizing wind-
generating fans is also recommended. Basedon the experiences reported here, this type of crosswind test
procedure appearsto offer significant promise for collecting driver-based subjective data of vehiclecrosswind
sensitivities. Whether such procedures can effectively replacenatural crosswind testing as a reliable method for
collecting driver subjectiveinformat ion remains to be seen.
References :
[1]. Vehicle Crosswind Stability. Chrysler Challenge Fund Project - #2000533, UMTRI, Sponsored by the Chrysler
Motors Corporation, 1987-1990.
[2]. W. Hucho, Aerodynamics of Road Vehicles. Butterworths, 1987.
[3]. A. J. Scibor-Ry lski, Road Vehicle Aerodynamics. Revised by D.M. Sykes ed. Wiley, New Yo rk, 1984.
[4]. W. Kamm, Anforderungen an Kraft wagen bei Dauerfahrten. ZVDI. vo l77, 1933.
[5]. I. Kohri and T. Kataoka, Research on Improvement of Cross -Wind Properties of Passenger Cars. JSAEReview,
Vo l. 10, No. 3, 1989. pp. 46-51.
[6]. H. Noguchi, Crosswind Stability and Suspension Properties. MIRA Translation by G. Wood, Rept No.
Translation 2/87. 1987.
[7]. T. Kobayashi and K. Kitoh, Cross -Wind Effects and the Dynamics of Light Cars. Int. J. of Vehicle Design.
Vo l. No. Special Publicat ion SP3, 1983. pp. 142-157.
[8]. J. P. Smith. Wind Gusts Measured on High-Speed Roads. The Motor Industry Research Association, Rept No.
197217, May, 1972.
[9]. V. T. Tran, Determining the Wind Forces and Moments Acting on Vehicles by Means of Pressure Transducers.
SAE Paper No. 900313, 1990.
[10]. J. D. Pointer, On-Road Calibration of the Dynamic Pressure Vector System. 10/ 17/ 88. Memo randum to G.
Ro mberg. Chrysler Motors Corporation. 1988.
[11]. M. Sayers, C. C. MacAdam and Y. Guy, Chrysler/UMTRI Wind -Steer Vehicle Simulat ion. User's Manual,
Version 1.0, vols I and 11, Report No. UMTRI-89-811-2, 1989.
[12]. L. Segel, On the Lateral Stability and Control of the Automobile as Influenced by the Dynamics of the Steering
System. Journal Engineering for Industry, Vol. 66, No. 8, 1966. pp.
[13]. R. H. Klein and H. R. Jex, Develop ment and Calibrat ion of an Aerodynamic Disturbance Test Facility. SAE
Paper No. 800143, 1980.
||Issn 2250-3005(online)|| ||January || 2013 Page 100